Introduction.

Sweet potato (Ipomoea batatas (L) Lam.) is sensitive to weed competition, and chemical control in Costa Rica is challenging due to limited local information on herbicide selectivity. Objective. To determine the selectivity of six herbicides applied at different stages of development to Criollo sweet potatoes and their effect on coverage, biomass, and crop yield. Materials and methods. Between August and December 2021, three experiments were conducted at the Fabio Baudrit Moreno Agricultural Experimental Station, Universidad de Costa Rica, Alajuela, Costa Rica. Herbicides were applied (in kg a.i. ha-1) as follows: a) Experiment 1: 0.36 of metribuzin, 1.50 of pendimethalin, 0.48 of clomazone, 0.075 of halosulfuron, 1.00 of linuron, and 1.25 of metolachlor, applied six days after planting (dap) in a randomized block design with four repetitions; b) Experiment 2: 1.00 of linuron and 0.36 of metribuzin 0.36, applied at -1, 15 and 30 dap, in a randomized block design with a 2x3 factorial arrangement and three repetitions; and c) Experiment 3: 0.075 of halosulfuron and 0.36 of metribuzin, mixed with 0.125 of fluazifop-p-butyl in a randomized block design with three repetitions. Weeding was performed at 15 and 30 dap in all cases. Results. Root yield was not affected in any of the three experiments. Experiment 1: Only halosulfuron caused mild damage to the plants; no herbicide affected above-ground biomass production. Experiment 2: Only linuron and metribuzin applied at 15 and 30 dap caused mild damage. Experiment 3: The mixtures caused temporary damage to the plants. Conclusions. The evaluated herbicides were selective to the Criollo sweet potatoes, and did not affect the yield of marketable storage roots. Only halosulfuron caused mild damage in all application forms, and metribuzin and linuron only in late applications.